SPECIFIC PERFORMANCE VERSUS DAMAGES FOR BREACH OF CONTRACT: AN ECONOMIC ANALYSIS

THE UNIVERSITY OF MICHIGAN LAW SCHOOL The Law and Economics Workshop Presents SPECIFIC PERFORMANCE VERSUS DAMAGES FOR BREACH OF CONTRACT: AN ECONOMIC ANALYSIS by Steven Shavell, Harvard THURSDAY, September 29 2005 3:40-5:15 Room 236 Hutchins Hall Additional hard copies of the paper are available in Room 972LR or available electronically at http://www.law.umich.edu/centersandprograms/olin/workshops.htm

Specific Performance versus Damages for Breach of Contract: An Economic Analysis Steven Shavell * Abstract: When would parties to a contract want performance to be specifically required, and when would they prefer payment of money damages to be the remedy for breach? This fundamental question is studied here and a novel answer is provided, based on a simple distinction between contracts to produce goods and contracts to convey property. Setting aside qualifications, the conclusion for breach of contracts to produce goods is that parties would tend to prefer the remedy of damages, essentially because of the problems that would be created under specific performance if production costs were high. In contrast, parties would often favor the remedy of specific performance for breach of contracts to convey property, in part because there can be no problems with production cost when property already exists. The conclusions reached shed light on the choices made between damages and specific performance under Anglo-American and under civil law systems, and they also suggest the desirability of certain changes in our legal doctrine. 1. Introduction When would parties to a contract want performance to be specifically required, and when would they prefer payment of money damages to be the remedy for breach? I study this fundamental question here and come to a conclusion based on a simple distinction between two types of contract: contracts to produce new goods or to provide services; 1 and contracts to convey existing goods or other property. 2 Setting aside qualifications, the conclusion that I will reach is that parties would tend to prefer the remedy of damages for breach of contracts to produce things, whereas they would often favor the remedy of specific performance for breach of contracts to convey property. 3 * Samuel R. Rosenthal Professor of Law and Economics, Harvard Law School. I thank Melvin Eisenberg, Allen Ferrell, James Gordley, Christine Jolls, Louis Kaplow, and Eric Posner for comments, David Katz and Gary Lee for research assistance, and the John M. Olin Center for Law, Economics, and Business at Harvard Law School for research support. 1 I will often refer to this category of agreements simply as contracts to produce things even though I mean to include contracts to provide services as well. 2 As will be discussed, I focus on contracts where the good or service is not readily available for purchase or sale on an organized market, but the latter case is discussed in subsection 2.4. 3 A different and, for some purposes, a better statement of the conclusion is that parties will tend to want damages to be the remedy when the reason for breach is high cost (as could only be true for a contract to produce) and would tend to prefer specific performance to be the remedy when the reason for breach is sale to an outside party (as could be true either for a contract to produce or for a contract to convey). 1

This conclusion will help us to understand the choices made between damages and specific performance under Anglo-American 4 and under civil law systems 5 and will suggest the desirability of certain changes in our legal doctrine. The conclusion and the analysis underlying it differ significantly from those in previous writing, as I will indicate after describing the organization and content of the article. I begin in section 2 with a theoretical, economically-oriented examination of damages and specific performance. 6 The question that I address there is what the parties themselves would want the remedy for breach to be. The point of departure for the analysis of this question is that contracting parties should in principle agree ex ante to choose the remedy that would maximize the joint value of the contract to them, where the joint value is the value gained by the parties less any expenses, costs of bargaining, and risk-associated disutility. The parties should want to maximize joint value essentially because, if a proposed remedy does not lead to the highest joint value, both parties can be made better off by agreeing to another remedy, generally after making a suitable price adjustment. If, for instance, they were contemplating specific performance but that remedy would lead to lower joint value than a damage measure, both the seller and the buyer can be made better off by changing from specific performance to the damage measure, after lowering the price to compensate the buyer if the buyer is made worse off because no longer guaranteed performance. I initially consider the choice of remedy in the context of contracts to produce (say, a contract to excavate a construction site). Here I explain that specific performance involves four disadvantages that would often lower joint contractual value. 7 First, sellers might have to perform even though that is very expensive (suppose that an excavator unexpectedly encounters hard rock) and outweighs its value to the buyer. Of course, in such circumstances, sellers might also negotiate for their release, but that would involve bargaining costs and might not result in an agreement. Second, the prospect of these problems associated with high production expense might lead sellers to take wasteful avoidance steps (to purchase rock-crushing machines even though that is intrinsically uneconomic). Third, the possibility of sellers having to pay large amounts for releases if performance would be very expensive (or worse, of actually having to perform) constitutes an undesirable risk for sellers. That is, even if excessive performance never 4 See, e.g., JOHN D. CALAMARI & JOSEPH M. PERILLO, THE LAW OF CONTRACTS ch. 16 (4 th ed., 1998), E. ALLAN FARNSWORTH, CONTRACTS ch. 12 B (3 rd ed., 1999), and EDWARD YORIO, CONTRACT ENFORCEMENT: SPECIFIC ENFORCEMENT AND INJUNCTIONS (1989). 5 See, e.g., John P. Dawson, Specific Performance in France and Germany, 57 MICH. L. REV. 495 (1959), G. H. TREITEL, REMEDIES FOR BREACH OF CONTRACT, VII INTERNATIONAL ENCYCLOPEDIA OF COMPARATIVE LAW 3-24 (1976), and KONRAD ZWEIGERT & HEIN KÖTZ, INTRODUCTION TO COMPARATIVE LAW ch. 35 (Tony Weir, trans., 3 rd rev. ed., 1998). 6 The analysis is economic in the sense that it is a systematic consideration of how parties would be expected to behave in the face of legal and other incentives and that it sometimes makes use of numerical examples. I do not believe that readers will have any difficulty following it. 7 The first and fourth of the disadvantages are well appreciated, whereas the second and third have not been emphasized. For further discussion, see infra subsection 4.3. 2

occurs because sellers negotiate releases, sellers bear large risk, a form of cost. 8 Fourth, the process of enforcement of an obligation to perform might involve substantial expense and result in subpar outcomes. These joint value-lowering disadvantages of specific performance generally would not arise under the expectation measure of damages, which is assumed to be the measure of damages in the analysis. Under the expectation measure, if it were very expensive to perform, sellers could, and usually would, breach and pay damages rather than perform (an excavator who encountered hard rock presumably would do this). Thus, sellers would not be forced to perform and ordinarily would avoid more than modest bargaining costs, would not be induced to spend wastefully on avoidance steps, and would not bear risk beyond that of expectation damages. Moreover, the parties would not bear the costs of enforcing specific performance. Next consider contracts to convey property. The uncertainty faced by a seller for such contracts concerns bids for the property (say a parcel of land) that outside parties might make. Importantly, production cost uncertainty would not be at issue, since by assumption the property to be conveyed already exists. Because the nature of the uncertainty is different for contracts to convey property, specific performance involves none of the disadvantages just reviewed for contracts to produce. 9 Sellers obviously do not have to perform or purchase a release when performance would be expensive, or bear associated risks for sellers can easily comply with specific performance by conveying the property that is by hypothesis in their possession. Moreover, a seller s performing when there is a high outside bid generally does not imply that the contracting parties would forgo a valuable opportunity for profit, since the usual presumption would be that the contract buyer could also sell to an outside bidder (an outside bidder who wanted a parcel of land could purchase it from the contract buyer as well as from the contract seller). Finally, enforcing specific performance of a contract to convey is ordinarily relatively straightforward and does not involve the difficulties often associated with mandated performance of a contract to produce something. What has just been stated is that specific performance does not lead to losses in joint value for contracts to convey property, but it does not suggest that specific performance would be superior for the parties to use of the expectation measure. Why might specific performance be superior, which is to say, why might use of the expectation measure lead to losses in joint value for the parties? The answer that is developed is that there is a danger of joint losses if the value of performance is underestimated. For in that event the seller might breach and sell to an outside bidder at a price below the value of the property to the contract buyer. There are then two possibilities. On one hand, the property might remain with the outside bidder, in which case the parties will clearly lose, as the property will have been sold at a price below the buyer s value. On the other hand, the contract buyer might purchase the property from the outside bidder. But in that case the buyer will generally have to pay more than the outsider had paid the seller, so that the outsider s profit will constitute a loss for the contracting parties. Under specific 8 The bearing of risk is a cost for parties who are risk-averse in the parlance of economics; see infra subsection 2.1. 9 This contrast is a central point of the present article. 3

performance, in contrast, contract buyers will never sell property to outside bidders unless the amount that they receive exceeds their valuation. This reason explaining why specific performance may be superior to the expectation measure applies to the degree that there is a chance that the expectation would be underestimated. 10 In section 3 of the article, I review the contours of the use of specific performance versus damage measures in Anglo-American, French, and German contract law and relate the choices made to the analysis of section 2. Under Anglo-American law, as readers know, specific performance is an exceptional remedy, employed mainly for contracts to convey property with unique or hard-to-evaluate aspects, but occasionally for contracts to produce things. It will be suggested that this pattern is broadly consistent with the theoretical analysis, especially because of the use of specific performance for contracts to convey property where the problem of joint loss to the parties under the expectation measure might be significant. However, it will also be suggested that the need for the inadequacy-of-damages test for use of specific performance for contracts to convey property is not apparent, and that it might be desirable to more widely grant specific performance for this class of contract. Under the French Civil Code, the remedy for contracts to produce things is damages, whereas the remedy for contracts to convey property is specific performance, so the distinction drawn in the Code is precisely the one drawn in the theoretical analysis here. French courts, though, sometimes employ a type of penalty to achieve effective specific enforcement of contracts to produce things. Under German law, specific performance is generally available as the remedy for breach of contract, except for personal service contracts. I do not reach any confident conclusion that one or another legal system is superior to the other in how it decides between specific performance and damages, but a tentative evaluation is that the German system is least consistent with the mutual interests of contracting parties. In any case, it will be interesting to observe that the three legal systems resemble each other in important qualitative respects, despite their nominal differences and their having different legal histories and rationales for their legal policies. A helpful way to understand why there is the resemblance that there is among the systems is through the lens of the mutual desirability of remedies, as revealed by the theoretical analysis of section 2. In section 4, I comment on major themes of thinking about specific performance versus damage payments for breach. Notably, I examine the view that there is a moral duty to obey a contract and thus a general reason in principle for specific performance to be the remedy for breach. I suggest that this view is misleading, as it does not reflect the fact that contracts often, if not typically, fail to provide explicitly for the particular contingencies that lead to breach, and thus that contracts are not natural to regard as promises for which a moral duty would attach. Indeed, I explain that, under the expectation measure, breach tends to occur exactly when parties would have agreed on 10 These paragraphs summarizing the theoretical conclusions reached in section 2 describe only central tendencies. As is discussed, the conclusions about the mutually preferred remedy may change if various assumptions are relaxed. For example, if it is assumed that outside bids are made only to contract sellers (not to buyers as well), the parties to a contract to convey property might prefer the expectation measure to specific performance; see infra subsection 2.4 on this issue. 4

nonperformance in a detailed contract with express provisions for all contingencies. Hence, there is a sense in which the conception that a contracting party has a moral obligation to perform when the party would want to commit breach and pay expectation damages is mistaken at a fundamental level. I next consider the familiar idea that inadequacy of damages as compensation for breach provides a rationale for use of specific performance. I stress, though, that it is not immediately clear why inadequacy of damages per se should lead contracting parties to want specific performance. For if a promisee knows that damages would not make him whole, he presumably could be compensated in advance for this disadvantage by an appropriate reduction in the contract price. 11 Thus, the notion that inadequacy of damages is a general problem for contracting parties, arguing for specific performance, seems mistaken. Still, for the reason that I sketched above, inadequacy of damages does lead to a joint problem for the parties that is answered by specific performance, but that logic is special to the context of contracts to convey property. I then discuss the relationship between this article and previous economically oriented writing on specific performance versus damages. With regard to production contracts, I note that several of the points that I mention are well recognized: excessively costly performance may occur under specific performance but not under the expectation measure; bargaining can often ameliorate the problem of excessive performance under specific performance, although at a cost; and specific performance may be difficult to enforce. What I add is that, even if bargaining is costless and always prevents excessive performance, specific performance still presents two major problems: it imposes risk on sellers because sellers have to pay for their release (in principle an amount up to their production cost); and it also motivates sellers to spend wastefully to avoid situations where production cost would be high and they could be held up by buyers. The major contribution of this article, however, flows from its observation that the comparison of specific performance to expectation damages changes its entire nature for contracts to convey property. As I stated, none of the problems afflicting specific performance where there is production cost risk apply where property is to be conveyed. Moreover, and importantly, I supply a new argument explaining why specific performance is jointly superior to expectation damages when such damages might be underestimated and property is to be conveyed. I distinguish this argument from one offered by Anthony Kronman, which I find unappealing because it is based on an assumption that buyers beliefs are systematically different from sellers. 12 I also discuss two articles, by Alan Schwartz and by Thomas Ulen, who recommend that specific performance be made routinely available as a remedy for 11 Compare, for example, the situation of a promisee who faces a twenty per cent likelihood of breach and who would receive damages equal to his losses to the situation of the promisee if the damages he would receive would be $10,000 less than his losses. If in the latter situation the promisee pays a price that is approximately $2,000 lower, he should be approximately as well off as if he does not pay a lower price but instead would receive higher damages. 12 Anthony T. Kronman, Specific Performance, 45 U. CHI. L. REV. 351 (1978). 5

breach. 13 I find their proposal to be inadvisable for the basic reason that they do not consider adequately the problems of specific performance for contracts to produce things. Another article on which I comment is by Melvin Eisenberg. 14 Additionally, I briefly discuss evidence that we have about parties preferences concerning the remedy for breach. Although this evidence is quite limited, it has some value and it does not lead one to suppose that parties have a strong desire for use of specific performance in domains where it is not now employed. 2. Theoretical Analysis 2.1 Assumptions and framework of analysis In the theoretical analysis presented in this section, I will be concerned with two stylized contractual contexts. As indicated in the Introduction, one is where the seller contracts to produce a good; the other is where the seller contracts to convey property. I will emphasize situations where the good to be produced or the property to be conveyed cannot be obtained on an organized market, in other words, where there is something particular about the good or the property that is the subject of the contract that distinguishes it from what could readily be purchased on a market. 15 In both contractual contexts, I consider uncertainties that might lead the seller to want to commit breach. In particular, I will assume that for contracts to produce things, uncertainty exists about production cost, so that if the cost would turn out to be high, the seller might want to commit breach. I will suppose that for contracts to convey property, uncertainty exists about the bids that outsiders would make, so that if an outsider s bid would turn out to be high, the seller might want to commit breach and sell to the outsider. 16 In reality, uncertainty about bids from outsiders exists for contracts to produce things as well (after a contract to produce a new good is made, the seller could encounter an outsider who makes a bid). It will be expositionally convenient, however, to discuss this case only after proceeding through the analysis under the assumption that the sole uncertainty affecting contracts to produce is production cost uncertainty. 17 An issue that arises if problematic contingencies occur concerns post-contractual bargaining. One assumption that I will make is that there is no post-contractual bargaining; the seller simply commits breach or not, based on his self-interest and the remedy for breach. The alternative assumption that will be examined is that there is postcontractual bargaining, and notably, that the seller might negotiate for release from an obligation to perform. The assumption of no post-contractual bargaining might fit some circumstances, for a seller might need to make a decision on the spot and not be in 13 Alan Schwartz, The Case for Specific Performance, 89 YALE L. J. 271 (1979) and Thomas S. Ulen, The Efficiency of Specific Performance: Toward a Unified Theory of Contract Remedies, 83 MICH. L. REV. 341 (1984). 14 Melvin A. Eisenberg, Actual and Virtual Specific Performance, 93 CAL. L. REV. 975 (2005). 15 See infra subsection 2.4, however, where I discuss cover. 16 For simplicity, I will suppose that there is no uncertainty about the buyer s value from performance, so that the buyer would not want to commit breach. 17 The case of both types of uncertainty for contracts to produce is considered in subsection 2.4. 6

immediate contact with the buyer or might find the cost of bargaining too large to justify incurring. In many, circumstances, however, post-contractual bargaining would be plausible. I will assume generally that such bargaining involves costs. I will also assume that bargaining might not succeed even though a mutually beneficial agreement exists in principle; asymmetry of information between the bargaining parties may lead them to misgauge one another and to an impasse. 18 In order to focus on the choice between specific performance and damage remedies, I will assume that a contract is of a very simple character: it names an unconditional duty to produce a good, or to convey property, as the case may be a price, and a remedy for breach either specific performance or expectation damages. A number of comments about these assumptions are worth making. First, that the contractual duty is unconditional means that the contract does not provide for parties to be excused from the obligation to perform under problematic contingencies, such as high cost. A contract that provided explicitly for all contingencies is a contract that the parties would want specifically enforced, since by definition it would reflect the true wishes of the parties, whatever might happen, and would say not only when there should be performance but also when there would be no duty to perform. In reality, of course, contracts usually mention some contingencies but still remain substantially incomplete; they do not provide expressly for many possible circumstances because of the impracticalities and costs that would be associated with making highly detailed contracts. The most convenient way to study the implications of incompleteness of contracts is to assume, as is done here, that contracts contain no contingent provisions. (If instead, I assumed that there were some contingent provisions, but not a complete set of contingent provisions, I would arrive at essentially the same qualitative conclusions that I reach below.) I will return to these points later, in section 4, when I discuss the view that there is a moral obligation to obey a contract, for the key to understanding the relevance of this ethical duty lies in recognizing that contracts are in fact incomplete. Second, by the remedy of specific performance, I mean the remedy that assures that the contractual duty is performed. The interpretation of this remedy depends on the contractual context and will be of relevance for some of the issues considered below. In the context of contracts to produce things, specific performance might mean forcing the seller personally to perform. Such literal performance would be needed if the seller were the only party who could perform, for instance, if the seller were a well known entertainer for whom there is no real substitute. Another possibility is that specific performance is equivalent performance, accomplished by having the seller arrange for, or pay for, a covering contract, whereby another party performs the stipulated contractual duty. That would be feasible if the seller were not the only party who could perform, for instance, if the seller were providing a common service, such as plumbing or electrical work, or if the seller were producing a fairly standard good. In the context of contracts to convey property, specific performance might be literal, meaning that the seller would be forced to convey the very property in his or her possession, such as a parcel of land or a 18 On asymmetric information and bargaining, see John Kennan & Robert Wilson, Bargaining with Private Information, 42 J. ECON. LIT. 343 (1993), ERIC RASMUSEN, GAMES AND INFORMATION (3 rd ed., 2001), and STEVEN SHAVELL, FOUNDATIONS OF ECONOMIC ANALYSIS OF LAW 89-91 (2004). 7

painting. Specific performance might also be equivalent, whereby a covering contract is employed to obtain essentially identical property (say bushels of wheat) for the buyer. 19 Third, the measure of damages is taken to be the expectation measure, the value of performance (net of the contract price to be paid), because this is the favored, central measure of damages that legal systems employ. It will be evident from the analysis how other damage measures would compare to specific performance. 20 Before proceeding, let me state how parties will be presumed to evaluate their contractual situations from an ex ante standpoint. One assumption that we will study is that parties are risk neutral. A risk-neutral party evaluates a situation involving risk in terms of its probability-discounted or expected value. For instance, suppose that a riskneutral seller would incur costs of $1,000 with a fifty per cent chance and costs of $5,000 with a fifty per cent chance. The expected value of costs would therefore be 50%H$1,000 + 50%H$5,000 or $3,000, and the seller would treat the risky cost situation as if the costs were $3,000 for sure. An interpretation of an expected cost, such as the $3,000 figure, is that it is the average amount parties would incur were they to find themselves repeatedly facing the same uncertainty. 21 The assumption of risk neutrality is simplifying and conventional to employ; it is useful to consider because it captures the notion that an individual cares about both the likelihood of an outcome and its magnitude. 22 If parties are risk neutral, they will want to choose contractual terms that result in the highest joint expected value. To illustrate, suppose that the parties are contemplating a contract in which remedy R would be used for breach and from which the buyer s expected value would be 100 and the seller s expected value would be 150, so that the joint expected value would be 250. Suppose that they contemplate a change in the remedy to R', and that this remedy would lead to an expected value of 80 for the buyer and 220 for the seller, so that the joint expected value would be 300, which is higher. 23 Why would the parties be thought to agree to switch to the different remedy R'? The 19 The reader should not be distracted by my interpreting specific performance either as literal or as equivalent. What I call equivalent specific performance would probably not be considered specific performance in Anglo-American law, although in civil law countries, such as Germany, it might well be. See infra subsections 3.1-3.3. In any event, it should also be observed that, where a covering contract can be made, its cost is often different from expectation damages. A classic instance is where a covering contract is needed to complete a construction contract, and completion costs more than the value it adds. 20 It should be noted that specific performance itself might be viewed as a very high measure of damages, for if a substantial enough sum would have to be paid for breach, there would presumably be no breach (in the absence of the judgment proof problem). Hence, in strict logic, one might view this article as comparing expectation damages with a very high measure of damages. 21 Were the seller to face the described risk one hundred times, his costs would be $1,000 about fifty times and his costs would be $5,000 about fifty times. Hence, his total costs would be about $50,000 plus $250,000 or $300,000, meaning that his average cost would be about $300,000/100 or $3,000. 22 On risk-neutrality, see, for example, ROBERT S. PINDYCK & DANIEL L. RUBINFELD, MICROECONOMICS 155-160 (5 th ed., 2001). 23 The reason that the joint expected value is higher might be that performance occurs more often when its cost is less than its value, or that breach occurs more often when the cost of performance would exceed its value. 8

answer is that the seller could afford to reduce the price by enough to make the buyer better off, and still the seller would be better off. 24 In effect, if the contractual pie the parties have to divide would be increased by a change in the remedy for breach, there has to be a way to slice the pie (by means of a price adjustment) so that each has more pie to enjoy and thus is happier. We will also consider, and in parts emphasize, the often more realistic assumption that parties are risk averse, that they care not only about expected value but also about variability, and especially that they will want to avoid losing a significant amount. 25 If one or both parties is risk averse, the contract remedy that they would agree to choose might not be the one that results in the highest joint expected value, for it might leave a risk-averse party bearing substantial risk. They might prefer a remedy that sacrifices some joint expected value in order to reduce risk-bearing by a risk-averse party, but one can still view them as seeking to maximize a broader concept of joint value. 26 2.2 Contracts to produce Here, as stated above, I examine a situation where production cost is uncertain when the contract is made. After the contract is signed, but before production would commence, I assume that the production cost becomes known. At this juncture the seller might commit breach. We want to compare specific performance to the expectation measure of damages for breach by examining to what degree these remedies promote or detract from joint expected value and risk since, as just explained, this tells us which remedy the parties would want to adopt. Efficiency of performance. Let us initially consider whether performance tends to occur when and only when its value exceeds production cost, when performance is said to be efficient. Such performance maximizes the joint expected value of the contract (ignoring for the moment possible bargaining costs and also risk aversion). Under the expectation measure, performance will automatically occur exactly when that would be efficient, as is well recognized. 27 To illustrate, suppose that the value of performance is $100,000, that a price of $40,000 is to be paid at the time of 24 Under the contract with remedy R' and the old price, the buyer s expected value is 80 and the seller s expected value is 220. As the seller reduces the price, the buyer will be made better off and the seller will be made worse off. At some price reduction, the buyer s expected value will have risen to 110, at which point the seller s expected value will have fallen to 190 (since the sum of values under remedy R' is 300 the change in price does not affect the sum of values, it is a mere transfer between the parties). At this price reduction, both the buyer and the seller are better off than under the initial contract with remedy R (the buyer is better off since 110 exceeds 100, the seller is better off since 190 exceeds 150). A general proof that parties will always prefer to switch to a remedy that achieves a higher joint expected value can be given along essentially the lines of this example. 25 See, e.g., PINDYCK & RUBINFELD, supra note 22, at 155-160 and STEVEN SHAVELL, ECONOMIC ANALYSIS OF ACCIDENT LAW 186-192 (1987). 26 If one defines joint value as joint expected value minus some amount to take into account riskbearing by risk averse parties, then one can often still phrase parties objective as maximization of joint value. 27 See, e.g., RICHARD A. POSNER, ECONOMIC ANALYSIS OF LAW 119 (6 th ed., 2003) and SHAVELL supra note 18, at 343-348. 9

performance, 28 and that there are two possible levels of production cost, a normal level of $20,000, occurring with likelihood 90 per cent, and an unusual level of $200,000, occurring with probability 10 per cent. Under the expectation measure, the seller would have to pay damages of $60,000 for a breach (since performance would be worth $100,000 to the buyer but the price he would pay would be $40,000). Hence, the seller would be led to perform when the cost would be $20,000 (earning a profit of $20,000 is better than paying damages of $60,000), but in the unusual event that the cost would be $200,000, the seller would be led to commit breach (suffering a loss of $160,000 is worse than paying damages of $60,000). Hence, the seller would perform when and only when the production cost would be less than the buyer s value of $100,000, which is efficient. 29 Under specific performance, in contrast, the seller would always perform, assuming provisionally (until the next italicized subsection) that there would be no postcontractual negotiation. In particular the seller would perform when production cost would be $200,000, even though the buyer s value is $100,000 and performance is inefficient. Because specific performance would result in a lowering of joint expected value, risk-neutral parties would choose the expectation measure over specific performance, according to the general logic discussed above in subsection 2.1. It may be helpful to demonstrate this explicitly. Suppose that the parties are discussing a candidate contract with a price of, say, $40,000 and with specific performance as the remedy. We want to show that both parties would prefer to change the remedy to the expectation measure if an appropriate adjustment in the price is made. If specific performance is the remedy, the buyer s expected value is $60,000 (namely, $100,000 $40,000) and the seller s expected profit is $2,000 (that is, 90%H($40,000 $20,000) + 10%H($40,000 $200,000)). If the parties contemplate a switch to the expectation measure with no change in the price, the buyer s value would obviously be the same, $60,000 (since he either receives performance or a payment of $60,000). The seller, however, would be better off, since the seller would escape having to spend $200,000 when production cost would be high. The seller s expected profit would rise to $12,000 (for it would be 90%H($40,000 $20,000) + 10%H( $60,000)). Since the seller would be better off if the price is unchanged, the seller can afford to offer a reduction in the price so as to make the buyer affirmatively happy to switch to the expectation measure from specific performance. For instance, suppose that the price is reduced from $40,000 to $35,000. Then both parties will be better off than under the originally considered contract with specific performance, for the buyer s value will be $65,000 instead of $60,000 and the seller s expected profit 28 The assumption that the price is paid at performance rather than when the contract is made is not essential. If the price is paid at the outset, the expectation measure would equal the full value of performance rather than the difference between the value of performance and price, but breach would occur under the same circumstances and the comparison of remedies would be unaffected. 29 To verify this point algebraically, let V be the value of performance to the buyer, P the price, and C the cost of performance. Assuming that P is to be paid at the time of performance, damages for breach under the expectation measure are V P. Now if the seller performs, his profit is P C, whereas if he breaches his profit is (V P). Hence, the seller will perform if and only if P C > (V P), or equivalently, if and only if V > C. 10

will be $7,000 instead of $2,000. 30 Note that the sum of the two parties expected values under specific performance would be $62,000 (that is, $60,000 + $2,000) and would be $72,000 under the expectation measure (that is, $65,000 + $7,000); the $10,000 increase in the sum is due to the avoidance of wasteful performance 10 per cent of the time. 31 The preceding demonstration, it should be stressed, shows that the parties find it mutually desirable to use the expectation measure. The point is not that it is socially desirable for them to employ the expectation measure so as to avoid wasteful performance (although that is also true). The point is rather that the parties each selfishly prefer to employ the expectation measure. Renegotiation and the efficiency of performance. Suppose that we now relax the assumption that there is no post-contractual bargaining between the buyer and the seller and we examine the possibility that the parties would bargain when production cost becomes known to the seller. Assume too that such renegotiation would involve a cost. How does this alter the analysis? The possibility of renegotiation would make no difference it is a moot issue under the expectation measure, as there is no reason for the parties to bargain about performance under that remedy. If production cost would be $20,000, less than the value of performance, the parties obviously have no reason to bargain, for the seller will want to perform; and if the production cost would be $200,000, exceeding the value of performance, the seller would commit breach and pay damages, so the seller does not need to bargain for discharge from the obligation to perform. The possibility of renegotiation would clearly make a difference under specific performance, however, for when the production cost would exceed the value of performance, the seller would want to pay the buyer for release from having to perform. If the price were $40,000 and the production cost would be $200,000, the seller would want to purchase freedom from the obligation to perform; the seller would pay up to $160,000 for a release since he would lose that amount if he performed, and the buyer would accept any amount over the $60,000 that performance would be worth to him. If the parties bargain and reach an agreement for the seller s release, the waste due to inefficient performance, here $100,000 (namely, $200,000 $100,000) would be avoided. Still, the renegotiation process would absorb time and resources, so these bargaining costs would lower joint value. Moreover, the renegotiation might not lead to an agreement; a breakdown in bargaining could lead to specific enforcement of the contract, so that wasteful performance would occur. Common experience and the expansive literature on bargaining tell us that mutually beneficial bargains are not always 30 The buyer s value will be $100,000 $35,000 = $65,000. The seller s expected profit will be 90%H($35,000 $20,000) + 10%H( $65,000) = $7,000. This argument that both the buyer and the seller can be made better off by changing from specific performance to the expectation measure can be made whatever was the initially-discussed contract price. 31 The waste is $200,000 $100,000 or $100,000. Since this waste is incurred 10 per cent of the time under specific performance, the expected loss to the parties is $10,000. 11

struck, even though they exist, a major reason being that parties may have different information and misconstrue one another s situation in some way. 32 In sum, although the possibility of renegotiation lessens the disadvantage of specific performance, that remedy still tends to lead to losses in joint expected value, due to bargaining costs and potential bargaining failure. The parties would thus still be thought to elect the expectation measure over specific performance. Wasteful preventive expenditures. Let us next consider the issue of sellers engaging in wasteful preventive expenditures under specific performance, to avoid being held up by buyers when sellers face high production cost. Knowing that he might face a production cost of $200,000 and have an obligation to perform, and knowing too that the buyer could extract as much as this amount, net of price, in exchange for a release, the seller has a motive to take steps to ameliorate the losses he would suffer were $200,000 the production cost. The seller might, for example, be led to purchase equipment that would only be of aid were production cost to be high (like the rock crushing equipment mentioned in the introduction, of value only if an excavator ran into an unusual problem with hard rock). Suppose, for instance, that by spending $5,000 on such equipment, the seller s high production cost would fall from $200,000 to only $125,000. The seller might find the $5,000 expenditure worth making, since that would enhance his bargaining position with the buyer if the production cost were high the buyer s maximum demand in that event would fall by $75,000. 33 Yet such holdup-induced expenditure by the seller would really constitute a waste for the parties, since the joint value maximizing outcome is for production not to take place whenever the production cost would exceed $100,000, which $125,000 does. One way to express this point is to say that, even if bargaining were a costless process and always led to efficient outcomes to releases for sellers whenever the production cost would be high, exceeding $100,000 the holdup-induced expenditure of $5,000 might be made, lowering joint value. In other words, holdupinduced preventive expenditures constitute another reason why specific performance lowers joint value and why the parties would be thought to prefer expectation damages. Risk imposition. Specific performance also may impose a substantial risk on the seller in comparison to the expectation measure. Under specific performance, the seller is faced with the risk of bearing a cost potentially as high as the loss he would suffer were he to perform. If the production cost would be $200,000 and the price $40,000, so that the seller would lose $160,000 were he to perform, the seller might have to pay up to this amount for a release from the buyer, and if renegotiation were to fail, the seller would 32 If, for example, the buyer believes the seller s production cost would be $500,000, the buyer might demand, say, $250,000 for a release, an amount that the seller would refuse to pay, as it would exceed his true loss of $160,000. 33 To illustrate why the seller might spend $5,000, suppose that the seller would always conclude a successful agreement for release when production cost is high and would be forced to pay all of the gain from the release to the buyer. Assume that the contract price, paid at performance, would be as above, $40,000. Hence, if the seller does not spend the $5,000, he would pay $160,000 for a release with probability 10 per cent, and if he does spend the $5,000, he would pay only $85,000 (his loss were he to perform, since his cost would be $125,000) with that probability. Therefore, the $5,000 expenditure would yield an expected savings to the seller of 10%H$75,000 = $7,500, which is greater. The qualitative point of this note would not be altered if, instead of the buyer s obtaining all of the gain from a release, he would only obtain part of the gain it is still true that the seller would have an incentive to spend purely to improve his bargaining position, not to improve actual outcomes. 12

definitely have to bear this amount. Under the expectation measure, in contrast, the seller s risk is limited to the $60,000, the value of performance net of price to the buyer. The significance of the factor of risk imposition on the seller depends on the degree of risk aversion of the seller. An individual or a small business, for example, might be quite risk averse, whereas a large corporation not so. Also of relevance is the probability distribution of the cost of performance; in some situations, such as where the seller could purchase cover at a known price, the risk would be cabined; in others, that might not be possible, or the cost of cover might vary, so that the magnitude of the cost could be very high. Administrability. Last, let us consider the ability of courts to enforce specific performance versus expectation damages. To enforce specific performance, the court must ensure that the stipulated performance is accomplished, meaning that the court must be able to ascertain the quality of performance to guard against its being inadequate. In some circumstances, the task could be difficult (whether an opera singer performed up to her usual standards), in others not (whether a plumber installed a new heating system). Another potential problem is recalcitrance of the seller. This might be an issue if specific performance is literal (the opera singer is required to perform) but should be essentially moot if specific performance is accomplished by cover (the buyer in a plumbing contract could hire another plumber to install the heating system). To enforce expectation damages, courts do not have to assess and oversee the quality of performance, for by hypothesis there is no performance. But courts need to estimate the value of performance (if this is not named as liquidated damages in the contract) and then to collect that amount. One might expect that the costs and difficulties of enforcing specific performance would usually dominate those of enforcing the expectation measure of damages, especially where specific performance is literal. But cases where specific performance would be easier to enforce are probably not infrequent. 34 That it may be more difficult to enforce specific performance than the expectation measure is a point about courts, whereas what is relevant for us is the well-being of the parties. However, the problems faced by courts in enforcement will tend to affect the parties, especially since the parties are involved in the legal process and find it costly. Summary. What has been presented is a set of reasons suggesting that the parties would prefer expectation damages to specific performance as the remedy for breach, since use of specific performance would tend to lower joint value and impose risk on the seller relative to use of the expectation measure. In particular, under specific performance, when production cost would be high, the parties would tend to bear costs of renegotiation, and wasteful performance might occur. Additionally, the prospect of high production costs may induce sellers to take wasteful defensive measures. Also, the possibility of high production costs imposes risks on sellers whether or not they are able 34 Suppose, for example, that the contract is to construct custom cabinets for a home in an area where there is no organized market for the kind of work involved and that the court can fairly readily determine whether the job would has been satisfactorily accomplished. At the same time, suppose that the value of performance to the buyer would be hard to determine and would be contested by the two sides (the degree to which the buyer cares about having custom cabinets instead of ready-made ones is very difficult to ascertain). Then enforcement of the expectation measure might be more costly than enforcement of specific performance. 13

to bargain for release. Finally, problems of administrability may be encountered under specific performance that would not be experienced under the expectation measure, although this latter factor applies mainly where specific performance is literal rather than equivalent, accomplished by cover. 2.3 Contracts to convey property We now consider a situation where the seller promises to convey property to the buyer, such as land or a moveable like a painting. After the contract is made, but before the property is conveyed, an outside party might make a bid for the property. I will generally assume that the outside party can as easily make a bid to the contract buyer as to the contract seller. This assumption is natural to make since, if an outside party is interested in purchasing property, the party would usually be able to determine who had rights to it and make his bid to that individual. (But alternative assumptions are examined in subsection 2.4, after the main analysis is presented.) I now compare specific performance to the expectation measure of damages. Efficiency of performance. In the present context, the joint value maximizing or efficient outcome is for the property to be conveyed to, and retained by, the buyer if and only if his valuation of the property exceeds the amount the outsider would pay; if the outsider would pay more than the buyer s valuation, it would be efficient for the property to be sold to the outsider. Under the expectation measure, performance will automatically occur when that would be efficient. Suppose, for example, that the value of the property to the buyer is $100,000, that a price of $40,000 is to paid at the time of performance, and that there are two possible levels of the outside bid, $80,000 with probability 80 per cent and $150,000 with probability 20 per cent. 35 Under the expectation measure, the seller would have to pay damages of $60,000 for a breach. Therefore, the seller would convey the property to the contract buyer when the outside bid is $80,000 (since he would prefer to receive the contract price of $40,000 than to obtain $80,000 but pay $60,000 in damages, yielding only $20,000 on net), but the seller would breach and sell to the outsider when the outside bid is $150,000 (since the seller would prefer to receive $150,000 and pay $60,000 in damages, yielding $90,000 in profit, to performing and receiving only $40,000). These outcomes are efficient, since it is joint value maximizing for the property to wind up in the hands of the outsider if his bid is $150,000, as that exceeds the $100,000 value of the buyer, but not for the property to go to the outsider if his bid is $80,000. Under specific performance, outcomes will also be efficient, since we are assuming that the outsider can as easily make his bid to the contract buyer as to the contract seller. If the outside bid is $80,000, the contract buyer obviously will not accept the bid. But if the outside bid is $150,000, the contract buyer will sell the property to the outsider. In particular, specific performance does not lead to the possibility of inefficient performance; the opportunity of the parties to avail themselves of a high outside bid of $150,000 does not depend on whether specific performance is the remedy. 36 35 More generally, there might be a probability of no outside bid as well as probabilities of many other outside bids. 36 I discuss in subsection 2.4 infra how the argument is affected if outside bids might not be made as easily to the contract buyer as to the contract seller. 14